Aligning fMRIPrep and QSIPrep Outputs

,
21

Hi,

Thanks for your reply! I’ll try it as soon as I have the data processed.

Best regards,

Manuel

22

Hello @Steven,
My apologies for reviving this old discussion. I have a related question I am hoping to ask. I’m trying to align the output in fmriprep and Qsiprep in space MNI152NLin2009cAsym and I’m a little overwhelmed by the various spaces like native space, MNI space. What I have tried is:

  1. Run fmriprep with recon-all with default output space MNI152NLin2009cAsym.
  2. Run Qsiprep with previous recon-all output for preprocessing and reconstruction in default output space T1W.

In my understanding, what I need to do are:

  1. Rerun Qsiprep with specified output space MNI152NLin2009cAsym.
  2. Run the above python script to registration fmri bold to dwi.

And I also wonder if the output of recon-all needed to be modified given the recon-spec is mrtrix_multishell_msmt_ACT-hsvs. Because I see that in log/citation.md of Qsirecon, it mentions that FreeSurfer outputs were registered to the QSIPrep outputs.

Thank you very much!

23

Hi @taka2223

This is generally more difficult and not recommended compared to doing everything in T1w space. What is your use case for doing it in MNI?

QSIPrep does not have MNI outputs for DWI. Warping DWI images to MNI is difficult because you have to deal with rotating the gradient table based on the warp. If you can wait until the end of the pipeline, you can just register streamlines to MNI (DIPY : Docs 1.7.0 - Applying image-based deformations to streamlines), but still I recommend doing everything in T1w spcae.

What are you using the recon-all outputs for outside of fMRIPrep/QSIPrep pipelines that would require additional registration?

Best,
Steven

24

Thanks for your quick response!

My goal is to consturcut the connectivity matrix of both functinal network and structural network and use them for further machine-learning based analysis. And it seems that many atlases are provided in MNI.

I find that in document of qsiprep it mentions we can use --output-spaces template and maybe I missunderstand it.

I’m sorry that I use the recon-all outputs just by following part of instructions in doc,doc2. As the table in doc2 says that the recon-spec mrtrix_multishell_msmt_ACT-hsvs needs freesurfer input. :dizzy_face: :dizzy_face:

In short,I want to use fmriprep and qsiprep to construct SC and FC networks. Thank you for your guidance - I appreciate your patience as I’m a beginner and continue learning neuroscience fundamentals and will provide additional details when needed.

25

QSIRecon registers atlases to your DWI image. So, you can keep your fmriprep output in MNI space and use an MNI atlas (I recommend XCP_D as a postprocessing pipeline), and keep your DWI in subject-space and let QSIRecon take care of registering atlases to your data.

That flag is deprecated.

You do not need to do anything to recon-all outputs, QSIRecon takes care of all registrations.

No problem! But since the original question has been answered, please open a new thread for additional questions.

Best,
Steven

1 Like